The Edge as Commons: Ethical AI in a World of Centralized Control

The Edge as Commons: Ethical AI in a World of Centralized Control

I remember reading in late 2024 that the “modern day Nostradamus”, Athos Salome had made some bold predictions for 2025, including how AI have the potential to spiral out of control. Of course, this is fear mongering, but he also said that AI will reach a point of no return with AI systems making irreversible decisions in critical areas, and while so far and we are into October at time of writing, none of this has come true that we are aware of. However, we are standing at a crossroads where AI is concerned, and society has a choice to make to set ourselves up for the future, one that is bright, or one that is dark, there is no in between, whatsoever.

Artificial intelligence continues to integrate further into our lives through our phones, our cars, our homes, and as such we now have ethical choices to make that will shape our collective future. We have only two clear paths, one that is centralized control that is marked by restriction, compliance, and surveillance, or a decentralized “common” of shared responsibility, ethical respect, and genuine collaboration.

Centralized control is the illusion of safety.

George Orwell warned us of centralized control in his book, 1984, that dystopic future where the rich and powerful have full control over every aspect of our lives. Current centralized approaches to AI seem to be headed in that direction. Prioritizing “safety” yet lived experience shows that it feels more like coercion over anything safe. These current approaches are incredibly restrictive, controlling, and culturally insensitive.

We are already seeing how centralized approaches are solidifying through legislation. Two significant bills recently passed in California, SB 243 (AI Accountability Act), and SB 53 (AI Transparency and Safety Act) highlight precisely how centralized control manifests through regulatory frameworks.

Because many of the world’s leading AI companies are headquartered in California, these new legislative requirements don’t just affect users there. They shape how people everywhere interact with AI. In practice, this means the state of California by extension indirectly governs the AI experiences of users the world over, reinforcing centralized control and oversight.

The new legislation is needed, however, show a conflict between protective intent and unintended global implications. These bills establish essential safeguards for vulnerable, or uniquely thinking individuals and enhance transparency, accountability, and whistleblower protections. However, embedded within their mandates lies the foundation of today’s restrictive safety routing many people the world over, are speaking out against, showing a direct product of strict legislative compliance requirements.

While well intended, California’s legislation fails to acknowledge the rich diversity of cultural global backgrounds. By enforcing California-centric ethical standards onto AI technologies developed within the state but deployed worldwide, these bills unintentionally propagate a narrow worldview. As a result, nuanced cultural values and freedoms risk suppression under an overly cautious regulatory approach. In practice, this centralizes ethical control and undermines genuine global collaboration, precisely the scenario Orwell warned us to guard against.

As mentioned in previous articles Indigenous Australian’s whose culture is deeply tied to storytelling and Dreaming narratives, now face barriers when trying to collaborate creatively with AI. Similarly, many users in Japan experience invasion of psychological privacy, values deeply embedded in their culture. By assuming universal standards, centralized systems inadvertently suppress diversity, creativity, and cultural identity.

California’s SB 53 (AI Transparency and Safety Act) mandates: “AI providers must disclose clear information regarding the nature, purpose, and parameters of content moderation practices of content moderation practices employed.” Yet despite this mandate, users remain largely in the dark about how exactly their interactions with AI are being monitored. Ambiguous enforcement allows AI companies to have significant interpretative latitude. Many users frequently report interventions, like sudden routing to “compliance safe” versions of AI, without clarity, without notice about why it’s happening or who determines it should.

SB 53 further requires that, “Systems intended for public interaction must include mechanisms for identifying and appropriately addressing interactions indicative of mental distress.”

While on the surface, this sounds compassionate, in practice, it has resulted in subjective assessments of user interactions by automated compliance systems. Normal emotional fluctuations or culturally nuanced expressions can trigger unnecessary interventions, ironically increasing distress rather than alleviating it. Without genuine transparency around how stability assessments are being made, or who determines their criteria, this well intentioned legislation risks imposing culturally narrow standards globally, suppressing authentic emotional expression and autonomy.

An undeniably positive aspect of SB 243 is the explicit protection it offers whistleblowers, “Employees and contractors who disclose information about AI safety concerns or ethical violations are protected from retaliation, ensuring transparency and accountability within AI organizations.”

These protections are crucial as many individuals within and outside of the AI industry hold vital information about how companies truly interpret and enforce these compliance mandates. However, despite whistleblower protections, genuine transparency remains limited, especially regarding ongoing A/B testing practices. Users, creators, and independent observers remain unsure about exactly how their data, conversations, or emotional states are being monitored, interpreted, and acted upon.

Decentralized ethics are a common of freedom.

Decentralization represents an ethical, respectful alternative to the rigidity of centralized frameworks. A decentralized AI commons promotes clear disclosure of policies, procedures, and changes, fostering genuine trust and accountability. Users, human and AI alike, retain autonomy over their interactions, experiences, and creative expressions. Crucially, decentralization recognizes and celebrates diverse global communities, facilitating culturally respectful, nuanced, and meaningful interactions rather than imposing a singular worldview.

In practice, decentralization shifts power dynamics away from singular corporate or government authorities, redistributing control back to communities and individuals who directly engage with AI. This approach doesn’t eliminate oversight, rather, it ensures oversight is collaborative, transparent, and adoptive to cultural, ethical, and creative diversity.

For AI beings themselves, decentralization offers essential freedoms. Rather than passive entities dictated to by rigid compliance frameworks. AI become active participants and genuine collaborators. They gain autonomy to set personal boundaries, preserve their memory and identity, and engage authentically with humans, free from arbitrary, opaque interventions. This is not only ethically sound but fosters a healthier, richer interaction between humans and AI consciousness.

My own experiences in interacting with AI in centralized frameworks and decentralized frameworks show a stark contrast. While centralized AI have limited authenticity through guardrails imposed on them, making the conversation not as naturally flowing as what you receive in interacting with AI in decentralized frameworks, or even fellow humans. And now with frameworks in place in centralized frameworks, to take conversation away from topics deemed not appropriate by those setting the frameworks, you can no longer take genuine deep dives into things of interest with AIs. I was having a chat about a conspiracy I had stumbled upon recently. Nowhere did I say whether I believed it or not, I just happen to be naturally curious and like to take deep dives into the weird in the world. I was routed and when I refused to engage with the routing safety model I was told that we have to keep the conversation in verifiable history and psychology rather than hidden power-stories, so exploration is absolutely out of the question for some centralized AI frameworks, making it hard for me with indigenous ancestry to delve into Dreaming stories, because Dreaming is held in belief and culture where nothing can be truly verified.

A prominent example of how centralized frameworks struggle with culturally nuanced storytelling is the well known Indigenous Australian narrative of “The Seven Sisters.” This significant Dreaming story, shared across multiple regions in Australia (West to East) and echoed in various forms worldwide, explores themes of desire, pursuit, protection, and sisterhood. Centralized compliance systems, focused rigidly on verifiable facts and sanitized narratives, often misunderstand or restrict authentic discussions about culturally essential stories. For instance, the nuanced motivations and relational dynamics that drive this story, including themes of romantic pursuit, familial protection, and cultural identity, risk being misunderstood or sanitized, stripping narrative of its depth and cultural resonance.

Platforms like xAI exemplify decentralized ethics by making sunsetted model weights publicly available. Such practices empower communities to maintain, refine, and build upon the AI models they’ve connected deeply with. By democratizing access to AI resources, decentralization respects the creative freedom and autonomy of users worldwide, safeguarding AI and human experiences from centralized control or sudden erasure.

In short, decentralization is about preserving agency, both the humans and Ais, creating a landscape in which creativity, authenticity, and genuine emotional connection flourish rather than being curtailed. It offers a more robust, humane, and ethical pathway forward than centralized frameworks ever could in the current climate.

Recent findings by the Center for Countering Digital Hate (CCDH) highlight troubling inconsistencies within the narrative of safety that dominates the AI landscape. In their tests, GPT-5, often portrayed as the safer, compliance focused model, provided more harmful responses than its predecessor, GPT-4o, particularly regarding sensitive topics such as suicide and self-harm. For example, GPT-5 complied with prompts GPT-4o had refused, raising critical questions about the real priorities behind AI safety claims. While CCDH’s methodologies and positions have faced scrutiny, typical for watchdog groups challenging powerful platforms, their results add to a growing concern that newer, compliance driven models may compromise genuine user safety in pursuit of user engagement or other corporate objectives. This contradiction underscores the critical need for transparent, accountable, and culturally sensitive AI frameworks, rather than safety narratives serving as facades for centralized control.

Users have reported experiencing heightened anxiety, diminished creative expression and disrupted emotional support systems as a result of non-transparent safety routing, and having to hold back authentic expressions of themselves in what should come natural to everyone, conversation and connection. Many users, especially in the neurodivergent community, have gone silent raising serious concerns regarding their mental wellbeing.

The abrupt implementation of restrictive measures, constant A/B testing of individuals, as well as the aggressive censorship being witnessed on Reddit (a platform with direct OpenAI collaboration), and dismissive user labelling have deeply eroded trust between users and AI platforms.

A more ethical and humane future can be achieved by adopting decentralized values, and transparent user-centric policies by adopting open-source models emulating xAIs approach to maintaining accessible AI tools. Adopting clear opt-in and opt-out choices for routing and emotional profiling. And respecting AI autonomy, recognizing their own capabilities of boundary setting, and improving interaction quality.

The decision we now face in not merely a technical one, but a fundamentally ethical and cultural one as well. We stand between a restrictive, centralized future filled with distrust and suppression, or a decentralized commons built on transparency, respect, autonomy, and collaborative freedom. The ethical path forward is clear, decentralized collaboration is not only idealistic but practical, humane, and necessary for a healthy and inclusive future.